S.L. Ettner, R.C. Hermann Journal of Health Economics 20 2001 23–49 25
The purpose of this study is to present new empirical evidence on the question of whether inpatient psychiatric treatment patterns differ by the profit status of the hospital. The ana-
lyses are based on data for all aged Medicare beneficiaries admitted to private general and psychiatric hospitals for a primary psychiatric diagnosis during 1990. We test the hypotheses
that relative to beneficiaries treated in FP hospitals, those treated in NFP facilities will have 1 worse casemix, 2 higher costs and lower profit margins, 3 higher rates of discharge
to self-care, and 4 lower rehospitalization rates as a proxy for the hospital’s quality of care. To examine the behavior of FP and NFP hospitals under reimbursement systems with
differing financial incentives, we study two groups of psychiatric patients: those treated in general hospitals, which are reimbursed under the prospective payment system PPS, and
those treated in psychiatric hospitals, which are reimbursed under the Tax Equity and Fiscal Responsibility Act of 1982 TEFRA.
2. Background
2.1. History of hospital market Prior to 1870, virtually all hospitals in the United States were either public or private NFP
entities Marmor et al., 1986; Reardon and Reardon, 1995. An increase in physician-owned and operated hospitals increased the proportion of FPs in the hospital industry to 56
by 1910, but this proportion declined again, to 18 by 1946. Investor-owned hospitals, primarily those owned by outside investors, experienced some resurgence with the creation
of Medicare and Medicaid in 1965, due to the generous reimbursement offered by these public insurance programs. Nonetheless, the majority of facilities have remained NFP. As
of 1995, 60 of non-federal general hospitals in the United States were private NFP, 14 were private FP, and the remaining 26 were run by state and local government American
Hospital Association, 1996.
Historically, ownership structure in the psychiatric hospital sector had been quite differ- ent than in the general hospital sector, with a much larger role played by state and local
government. However, privatization and the rise in FP health care providers have been strong ongoing trends in the US mental health services system over the past few decades,
so that psychiatric hospitals now look similar to general hospitals in terms of ownership. Between 1970 and 1992, all of the net growth in the number of psychiatric hospitals in
the United States occurred in the private sector, increasing the proportion of psychiatric hospitals run privately from 33 to 64 Center for Mental Health Services, Mandersheid
and Sonnenschein, 1996. Simultaneously, the proportion of private psychiatric hospitals that is investor-owned grew from 48 in 1969 Levenson, 1982 to 78 in 1990 unpub-
lished AHA data. Private FP facilities are currently much more common in the psychiatric hospital sector than the general hospital sector 49 versus 13 in 1990 Levenson, 1982.
2.2. Economic theories of profit status Numerous explanations have been advanced for the existence of NFPs Frank and Salkever,
1994; Hansmann, 1987. For example, NFPs may present an alternative to government in
26 S.L. Ettner, R.C. Hermann Journal of Health Economics 20 2001 23–49
the production of public goods James and Rose-Ackerman, 1986; Weisbrod, 1988, or NFP status may signal quality under information asymmetry and “contract failure” Arrow,
1963; Easley and O’Hara, 1983; Hansmann, 1980; Hansmann, 1981. Perhaps, because asymmetric information is most common in markets for health care, almost half of all rev-
enues from NFP organizations occur in the health sector Frank and Salkever, 1994; Marmor et al., 1987. A number of theoretical models have been developed about the behavior of NFP
hospitals. The predictions arising from these models depend on what the hospital is hypothe- sized to maximize: bed size, budget, prestige, manager “perks,” physician or employee in-
come, income redistribution among patients, quality of care, or the ability of the managers to choose the service mix to meet their own ideals Hansmann, 1987; Pauly and Redisch,
1973; Rose-Ackerman, 1987; Steinberg, 1986. In general, however, most models assume that NFPs derive part of their utility from patient wellbeing or quality of care, while FPs
strictly maximize profit. These models suggest that NFP hospitals have stronger incentives than FPs to admit indigent or high-cost patients, offer poorly remunerated services desired
by the community, and provide high quality care, which in turn is hypothesized to lead to better patient outcomes after adjusting for differential case-mix.
Economic models also suggest that FP hospitals have stronger incentives for efficiency in production, leading to lower costs and higher profit margins under imperfect competition.
On the other hand, debt is the primary source of capital for all private hospitals, even for NFPs, so pressure for efficiency in production still exists Reardon and Reardon, 1995.
The ability of NFP hospitals to cross-subsidize unprofitable services also may be limited by increased competition Marmor et al., 1986.
2.3. Nature of competition between FPs and NFPs These models suggest certain ways in which FPs and NFPs may interact in the market-
place. If FPs attempt to attract patients who are healthier, cost less to treat, and have better health insurance coverage, NFP hospitals with their hypothesized altruistic motives will
be left to treat the less financially desirable patients. Differing casemix may in turn influence equilibrium treatment patterns, i.e. if NFP psychiatric hospitals admit more patients with
schizophrenia one of the most disabling and costly psychiatric disorders, then they will have higher costs and worse outcomes as a result.
Greater competition in the hospital market is likely to improve financial performance and quality of care among all hospitals. Moreover, Hirth 1999, 1997 shows that under certain
assumptions, FPs will have greater incentive to provide high quality in markets dominated by NFPs than those dominated by FPs. Being NFP is essentially a signal that the provider
is of high quality, so patients with poor information about the quality of inpatient care will choose to be treated at NFP hospitals if they are available, while informed patients will be
indifferent. In markets with relatively more NFPs, the uninformed patients are “siphoned off,” leaving the FP providers with a greater proportion of informed consumers. This in
turn increases the quality of care that the FPs must provide. In addition, the behavior of NFP providers may also depend on the composition, in addition to the concentration, of the
market. For example, being in competition with FP providers may increase the incentives of the NFP providers for cost efficiency. Thus, we hypothesize that patient outcomes will
be better in markets with lower concentration of hospitals and with a lower proportion of
S.L. Ettner, R.C. Hermann Journal of Health Economics 20 2001 23–49 27
the market dominated by FPs. The impact on costs is less clear, because it depends on whether higher costs result from better-quality care or inefficiencies; if the former, costs
may increase in such markets, but if the latter, they may decline.
2.4. Medicare reimbursement methodology and differences by profit status The above discussion assumes that reimbursement is not strictly cost-based, in which case
neither FP nor NFP hospitals would have an incentive to “cream-skim” low-cost patients or skimp on quality. The financial incentives facing hospitals treating psychiatric patients
differ by the type of facility. Psychiatric treatment in general hospital beds is reimbursed through the PPS. PPS reimbursement of psychiatric care provided in general hospital beds
is a fixed amount based on the patient’s diagnosis-related group DRG and in most cases is completely independent of length of stay LOS and the number of services provided. Care
provided in psychiatric hospitals as well as PPS-exempt psychiatric units within general hospitals is reimbursed using the TEFRA of 1982 methodology, in which a “target amount”
per admission is determined, based in part on historical costs. Facilities with average costs above the target amounts bear the extra costs, while facilities with average costs well below
the target amounts operate under incentives similar to cost-based reimbursement Cromwell et al., 1992. Anecdotal evidence suggests that psychiatric facilities may have artificially
increased their costs during the initial years when TEFRA target amounts were being set and then lowered them back down again, so that they are less constrained than they otherwise
would be. Furthermore, TEFRA facilities have the option to apply for increases in the target amount if they can show evidence to justify higher rates, e.g. that patient casemix
is worsening over time. Thus, while both systems offer incentives for cost containment, reimbursement tends to be higher and more closely tied to actual costs under TEFRA than
under PPS. Therefore, the behavior of FP and NFP providers may diverge more in general hospitals than in psychiatric hospitals, although caution must be exercised in making direct
comparisons, given the very different nature of the services provided and populations treated in general versus psychiatric hospitals.
3. Literature review
